Bridging or Bonding: Economic Payoff to SocialCapital, with an Application to Russia

Rinat Menyashev, Leonid PolishchukCenter for Institutional Studies at the Higher School of Economics,

Moscow

February 22, 2011

Abstract

Pro-social norms and other ingredients of social capital are shown to be con-ducive for economic development,institutional performance, and quality of gov-ernance. No such analyses were available for Russia so far, and the present paperfills this gap. We propose a model which differentiates the economic impact ofbridging and bonding social capital – while the former increases government ac-countability, the latter is mobilized to seek protection from government predationand make up for insufficient public provision of social services. We show that inequilibrium such grassroots private alternatives to publicly supplied institutionsand programs could have overall detrimental impact on development by reducingpolitical costs of abuse of power. These conclusions are confirmed empirically byusing data of a major Russia-wide survey held in 2007. We establish a significantpositive relationship between bridging social capital and urban development inRussia; bonding social capital has a strong negative impact on development

JEL codes: D7, R11Keywords: social cohesion, bridging and bonding social capital, government ac-

countability

1 Introduction

In the course of the last few decades there have been several major updates of economists’views of what factors contribute to economic growth and welfare. The conventionalgrowth theories dominant in 1950s-1970s emphasized the importance of investments inphysical and human capital. From 1980s onwards the emphasis has shifted on institu-tions, such as markets, contracts, property rights, rule of law, good governance etc. A

1

large body of evidence was presented in support of the claim that good institutions areindispensable for economic efficiency and factor accumulation. The “Institutions Rule”view (Rodrik, Subramanian, Trebbi, 2004) had straightforward policy implications –key to economic development is in institutional and policy reform.

Nonetheless in many instances institutional reforms in transition and developingcountries failed to deliver expected outcomes – the allocation of economic roles, powerand resources remained unaffected by institutional change (the resilience known as‘the invariance principle’ (Acemoglu, Robisnon, 2008)). Moreover, on some occasionsinstitutions and policies that were expected to improve welfare and facilitate growthhad the opposite effect, making matters worse (Putnam, 1993).

Such ‘surprises’ of institutional reform (Roland, 2000; Polterovich, 2007) highlighteda yet another important development resource – a ‘missing link’ (Guiso, Sapienza, Zin-gales, 2010) commonly known as social capital. Numerous definitions of social capitaloffered in the literature vary from broad views treating as social capital any sharednorms and values (Ostrom, 2000) and social structures (Dasgupta, 2003) to much morerestrictive interpretation which considers as social capital only virtuous characteristicsof the society that facilitate development and improve welfare (Fukuyama, 1997). Themiddle ground interpretation that does not render circular the relation between socialcapital and development outcome is the view of social capital as the capacity for self-organization and collective action in pursuit of some common good (Putnam, 1993;Woolcock, 1998). The main ingredient of such capacity is cohesion based on trust,social norms, values, and networks.

Social capital could be instrumental for economic development in two importantways reflecting two distinct patterns of collective action. First, it cuts transactioncosts in the private sector: trust and social connections facilitate investments and trade(Arrow, 1972), and self-organization offers private solutions of public problems. Second,social capital is indispensable in resolving the agency problem between governmentand society. Government accountability can only be ensured if there is sufficient civicculture (Almond, Verba, 1963) at the grassroots, i.e. the appreciation of political rightsand freedoms, awareness of public affairs, and the sense of civic duties and personalresponsibility for social well-being. These two mechanisms represent resp. horizontaland vertical ‘transmission channels’ between social capital and economic outcomes.

Social capital and formal institutions can be substitutes and/or complements. Tothe extent that social capital and institutions substitute for each other, they offerresp. formal and informal solutions of the same coordination problem, and henceone can expect that returns to social capital should be particularly high when formalinstitutions (and government-provided public goods and services) are in short supply orof poor quality (Durlauf, Fafchamps, 2005; Easterly, Ritzen, Woolcock, 2006). The flipside of this logic is that institutional reform remedies a lack of social capital (Knack,Keefer, 1997); indeed it is argued that extensive social programs of welfare state couldcould “crowd out” social capital (Wolfe, 1989). On the other hand without sufficient

2

social capital formal institutions and public policies could be either idled or capturedand subverted by narrow interests (Polishchuk, 2010), in which case institutional andgovernance reforms do not bring about desired results. In both cases, whether throughsubstitution or complementarity, social capital could be highly relevant for economicdevelopment and social welfare.

Such relevance has been empirically established in numerous publications at themacro-, meso-, and micro levels – for nations, regions, cities, local communities, aswell as for various public services and fields of social and economic activities. Thesestudies, while in general corroborating the view of social capital as a developmentresource, produce a more nuanced and complex picture – the impact of particularingredients, forms and types of social capital is highly context-specific. Thus, whatis known as ‘bonding’ social capital upholds collective action within narrow confinesof smaller isolated groups providing ‘club goods’ for group members. The impact ofsuch activities for broader social welfare could be detrimental, if smaller ‘Olson groups’are engaged in socially wasteful rent-seeking, or if they divert their resources andenergy from eliminating root causes of social and economic problems. On the otherhand, ‘bridging’ social capital facilitates the creation of broad society-wide coalitions(‘Putnam groups’) which advance social welfare by producing public goods, such asefficient public sector governance.

Economic payoff to social capital is measured in the literature by using nationalor regional data. In such studies various indicators of economic development, welfare,quality of institutions and governance are related to panoply of social capital mea-sures. Cross-country studies reveal tangible relations between economic outcomes andsocial capital; however, profound differences between countries, which are only partlycaptured by control variables, make such estimations less reliable. More appropriateframework for establishing an association between social capital and economic perfor-mance is provided by regional data within a given country. Such analyses have beenperformed for US states, as well as for regions of Germany, the UK, Poland, the Eu-ropean Union, China, etc. In most of these studies (regional) government efficiency,public service delivery and other outcomes are shown to be in statistically significantrelations to relevant measures of social capital.

No estimations of this sort have been obtained yet for Russia. While for mostother countries social capital is shown to have significant economic returns, there areconflicting arguments as to whether such payoff of comparable magnitude could beobserved in Russia as well.

First, there are doubts about the quantity and quality of social capital in Rus-sia. One of the most commonly used measures of social capital – the index of trustcalculated by using data of the World Values Survey – puts Russia and most othercountries of the former Soviet Union below the median among the nations covered bythe survey. Lack of trust and other ingredients of social capital in Russia is consistentwith the conjecture, posited by Putnam (1993) and later supported by an in-depth

3

econometric analysis for European countries and regions (Tabellini, 2008), that socialcapital accumulation is fostered by non-authoritarian political regimes. Furthermore,economic transition in the former Soviet Union and Central and Eastern Europe isshown to have significantly eroded social capital (Aghion et al., 2010), and such lossesin Russia were most profound across the post-communist region. According to Kumlin,Rothstein (2005), this could at least in part be due to an abrupt collapse of heretoforeuniversally available welfare-state programs, and of widely perceived injustice of state’sdealing with people. Quality-wise, Rose (1998) maintains that the existing stock of so-cial capital in Russia, low as it may be, is also obsolete and unsuitable to maintainmodern institutions of market democracy. This concern finds support in the interna-tional distribution of the aggregate index of ‘civic capital’ which puts Russia in thebottom quartile among 70 plus countries (Guiso, Sapienza, Zingales, 2010). Last butnot least, high degree of centralization of economic and political life in contemporaryRussia (known as the ‘vertical power’) leaves little space for grassroots initiative andself-organization.

All of the above makes one to expect that evidences of economic payoff to socialcapital in Russia would be hard to find, as social capital in the country is likely insuf-ficient and/or ‘idled’ by the political system and excessive government control. Andyet according to the contrarian view, weakness of official institutions and lack of publicgoods supplied by the government in fact raises the returns to social capital whichprovides informal grassroots fixes of institutional and governance failures.

An empirical confirmation of the skepticism about economic significance of socialcapital in Russia would validate the commonly expressed opinion that the society can-not be a driving force in the country’s development and modernization. If howeverthose views are refuted, Russia can be considered ‘a normal country’ (Shleifer, Treis-man, 2005) where social capital can improve institutions and governance; in that casea development scenario in which the society plays an active role becomes possible.Measurement of economic payoff to social capital in Russia is therefore a matter of notjust academic, but also practical significance. It is also important to find out whatkinds of social capital, if any, could affect social and economic outcomes in Russia.

In this paper we propose a simple economic model that describes outcomes of bridg-ing and bonding forms of social capital for social welfare and public sector governance.Predictions of the model lead to hypotheses which are tested by using data from amajor survey conducted in Russia in 2007 as part of the “Geo-Rating” project. Linksbetween social capital and development are explored at the city level. Factor analysisreveals three forms of social capital – bridging, bonding, and civic culture, which arelatently present in the data. Stocks of social capital exhibit significant variations fromone city or town to the other; it can therefore be concluded that there are more andless “civic” cities in Russia. The observed variations are found to be associated withsocio-economic conditions in the city (town, village); moreover, bridging social capitaland civic culture advance local development, whereas bonding social capital retards it.

4

Russia therefore is a ‘normal country’, at least when it comes to the impact of socialcapital on economic outcomes: more civic cities are better-off than less civic ones.

It is further demonstrated that the main ‘transmission mechanism’ between socialcapital and economic outcomes is the performance of municipal governments, whichis significantly improved by bridging social capital and civic culture, and adverselyaffected by bonding social capital. It is noteworthy that such links cannot be estab-lished at the regional level; one possible explanation is the political difference betweenRussian cities and regions (oblasts, krajs, and republics) – city mayors are more oftenthan not electable, while regional governors since 2004 are federal appointees.

Causality between social capital and economic outcome is confirmed by using two-stage least squares regression analysis, where the size of the middle class is shown tobe a valid instrument for bridging social capital, thus confirming the role of the latteras an (urban) development factor and resource.

2 Social capital impact measurement

Earlier social capital studies were mostly qualitative by their nature and did not at-tempt to establish an empirically grounded relationship between social capital anddevelopment. Putnam’s famous book (1993) was the first scholarly work where payoffto social capital was supported by data: it was argued that higher stocks of socialcapital in the northern part of Italy allowed northern provinces to make full use of adevolution of power and resources from the central government to the regions, whereasinsufficiency of social capital in the South of Italy precipitated failure of the samereform.

Since Putnam (1993) measuring payoff to social capital has become a rapidly grow-ing ‘cottage industry’ in social science; the vast literature on the subject is reviewed byHalpern (2005); Durlauf, Fafchamps (2005); Guiso, Sapienza, Zingales (2010). In thefirst attempts to prove (and measure) the economic impact of social capital by meansof econometric analysis, Knack and Keefer (1997) and La Porta et al. (1997) usedcross-country data. In these papers rates of economic growth and measures of socialwelfare and government performance were dependent variables in regression models,while various social capital indexes served as independent variables. It was shown thattrust had positive statistically significant relations to economic outcomes (with causal-ity confirmed by appropriately chosen instrumental variables), whereas no such relationwas found for associational activities. This was an indication, repeatedly confirmed bysubsequent studies, that contrary to Putnam’s earlier expectations, social capital is nota generic “commodity” with all of its components invariably relevant in any develop-ment, organizational etc. context, and that the identification of types of social capitalthat are economically valuable under particular circumstances is a non-trivial problemand should be dealt with on a case-by-case basis.

5

Putnam’s pioneering work opened a strand of empirical research where social cap-ital’s impact was measured by using regional data. Knack (2002) established a sig-nificant impact of social capital on state governments’ performance in the US. It wasshown that trust in the society, volunteering, and indexes of civic maturity are goodpredictors of the quality of state government services and regulations. Associationalmembership was not found to be of economic significance, and an attempt to find suchrelations by differentiating between ‘Olson-like’ and ‘Putnam-like’ groups was unsuc-cessful. However, in a different study using US data association membership was shownto have tangible economic impact at the county level (Rupasingha, Goetz, Freshwater,2002).

Similar links have been confirmed for a number of other countries. Thus, in Chi-nese regions trust is highly correlated with population income, economic growth, invest-ments and the number of firms (Zhang, Ke, 2003). Casey (2004) established statisticallysignificant correlation between trust among individuals and in political institutions, onthe one hand, and bureaucratic efficiency, on the other, for British regions. At the sametime some other studies present less clear-cut pictures. Thus, for German regions thecontribution of social capital in its traditional interpretation is ‘fading’ in the shadowof more significant cultural factors, such as ‘market’ or ‘hierarchical’ values (Blume,Sack, 2008). In Poland social capital, measured by associational membership, is notfound to be making statistically significant contribution to economic growth and taxcollection in various administrative units of the country (Dzialek, 2009).

Measurement of economic payoff to social capital is closely related to the identifica-tion of social capital’s roots and origins – the latter could serve as causality-establishinginstruments for social capital. On a number of occasions religion and social homogene-ity were used as such instruments, but lately more popular choice was political history,based on Putnam’s conjecture that historic experience of democratic self-rule createssocial capital. This view is confirmed by Guiso, Sapienza, and Zingales (2008), whoshow that Italian cities and regions that were self-governed in the past have highersocial capital endowments than those that were under colonial rule, and that such dif-ferences are indeed valid instruments for social capital. Tabellini (2008) reaches thesame conclusion by using data for 69 European regions.

Studies of social capital’s outcomes in Russia so far have been more qualitativethan quantitative and not sufficiently comprehensive. Petro (2001) argues that greatersuccess of economic reforms in Novgorod region was due to higher social capital stockmeasured by association membership and civic initiative participation. Marsh (2000)calculates a ‘civic society index’ for Russian regions, which is shown to be positivelycorrelated with political engagement of population; however no attempt was made toestimate economic payoff to the so measured social capital. Kennedy and Kawachi(1998) found a link between the insufficiency of social capital and steep increase inmortality observed in Russia in the first decade of marker reforms; this is consistent withthe robust relationship, observed in other countries, between social capital and physical

6

and mental health (Halpern, 2005). A more recent study (Eberstadt, 2010) concludesthat social capital deficiency could be one of main causes of the present demographiccrisis in Russia, which is a ‘negative’ confirmation of social capital’s significance for thecountry’s social and economic outcomes.

The reviewed literature demonstrates that measurement of economic payoff to socialcapital and cohesion is a complex but realistic task. Its complexity is in part due tomultiplicity of social capital’s interpretations and meanings (Woolcock, Narayan, 2002,Durlauf, 2002, Guiso, Sapienza, Zingales, 2010) and well-known difficulties of accuratemeasurement of trust (see e.g. Glaeser et al., 2000), values, associational activitiesetc. Nonetheless more often than not the impact of cohesion and capacity for self-organization for economic outcomes can be empirically confirmed, and Russia, as weshow below, is no exception to this pattern.

3 The model

Modeling social capital’s impact on economic outcomes is still in its infancy, and nosufficiently universal and encompassing approaches have been developed so far. Zakand Knack (2002) offer a model that captures trust’s beneficial impact for investments;this model therefore deals with the ‘horizontal’ transmission channel for social capital.Weingast (1997) uses game theory to demonstrate importance of social capital forsustaining democracy, rule of law, and limited government. Glaeser et al. (2002) modelindividual investments in social capital in conjunction with externalities and networkeffects. Tabellini (2008) explores bilateral links between institutions and culture; hisanalysis demonstrates how trust creates grassroots demand for good institutions andgovernance, which in their turn facilitate productive economic activities and suppressrent-seeking. Polishchuk (2008) uses an economic model to investigate the role of socialcapital in the working of corporate social responsibility. Aghion et al. (2010) presenta model where trust and values in the society are related to the scale and scope ofgovernment’s presence in the economy and the quality of government regulations.

The stylized model that follows is concerned with the vertical ‘transmission chan-nel’ of social capital, whereby the latter’s role is to improve government accountability.The model is custom-built to capture the impact of different kinds of social capital.It incorporates Weingast’s (1997) idea that good governance ensues when sufficientlylarge social coalitions defend their economic and political rights against possible ex-propriation by the government when it ‘transgresses’ its constitutional boundaries andotherwise abuses power (see also Kuran, 1991). Effectiveness of such actions requiresbridging social capital and civic culture. Cohesion-based bridging social capital isneeded so that coalitions of sufficient size acting in public interest could be formed andsustained. The role of civic culture is to put government accountability on such coali-tions’ agenda; to this end, good governance should be perceived by coalition members

7

as a matter of high importance and personal responsibility.In the model the bonding form of social capital is mobilized to mitigate the damage

caused by government’s malfeasance, rather than stopping such malfeasance in thefirst instance. The relief is achieved locally and based on cohesion limited to smallergroups, and materializes in the form of club goods that substitute for insufficient pubicgoods supplied by the government, or in the form of shielding group members fromgovernment abuse. Government accountability that precludes transgression cannot bean objective of such groups which are too small, isolated and dispersed for the taskand do not have such matters on their agendas.

One should expect positive economic payoff to bridging social capital and civicculture through improved public sector governance. The impact of bonding socialcapital on socio-economic outcomes is a priory ambiguous: on the one hand it has apositive direct effect by improving the lot of small groups’ members who obtain relieffrom government abuse, but on the other hand it indirectly encourages greater abuseby lowering its economic and hence political costs.

We follow the tradition in the political economy literature (see e.g. Grossman,Helpman, 2001) to model imperfect government accountability by assuming that thegovernment maximizes a weighted sum WG+aWS of its own immediate economic wel-fare WG and the aggregate welfare WS of the rest of society (private sector); here themultiplier a ∈ [0, 1] represents the degree of government accountability. In what followsthis multiplier is an aggregate of bridging social capital and civic culture, as both ofthese ingredients are required for accountable governance.1

Suppose that the government abuses power in order to extract and appropriateincome D ≤ D from the society (private sector); D represents physical, institutionaletc. limits to such expropriation. The private sector comprises a unit continuum ofagents, and government’s action causes each agent a material loss C0(D) ; however ifan agent is a member of an organized group that seeks collective grassroots protectionfrom government abuse, these losses are reduced to C1(D). The width of this grassrootsprotection depends on the stock of bonding social capital which is measured by theshare w ∈ [0, 1] of agents organized in such groups.2 We assume that

D ≤ C1(D) ≤ C0(D),∀D ≥ 0. (1)

(the first of these inequalities implies that grassroots protection can at best elimi-nate excess burden C0(D)−D of government abuse), and that the functions C0(D), C1(D),and C0(D)− C1(D) are all monotonically non-decreasing. If W0 is the aggregate wel-

1In a more detailed version of the model bridging social capital and civic culture are presentexplicitly and separately from each other; results of such model’s analysis remain qualitatively thesame.

2One can think of group formation as random events in which case w is the expected share of agentsorganized in such groups; alternatively bonding social capital could be confined to certain parts ofsociety, in which case w is the share of such parts.

8

fare of the private sector before government transgression, then after the transgressionprivate sector welfare is reduced to

WS = W0 − wC1(D)− (1− w)C0(D); (2)

and assuming WG = D, the expropriated income can be found from the followingproblem:

maxD

[D − a(wC1(D)− (1− w)C0(D))]. (3)

Comparative statics analysis of the above problem leads to the following conclusion.

Proposition 1 The expropriated income D = D(a, w) is (non-strictly) increasing inw and decreasing in a.

Proof. According to the “supermodularity lemma”, the solution of the problem

maxx

f(x) + ag(x)

is monotonically non-decreasing in a as long as the function g(x) is monotonicallyincreasing. To establish the required property of D(w) as a function of w, the govern-ment’s objective function should be rearranged as D− aC0(D) + aw(C0(D)−C1(D)),

and of a – as1

aD − (wC1(D)− (1− w)C0(D)). �

The above analysis of the model shows that broad-based social cohesion and civicculture work through government accountability to restrict possible abuse of powerand thus improve the quality of governance and formal institutions. On the contrary,narrowly-based social cohesion is detrimental for the quality of governance as it makesthe society more resilient and hence tolerant to abuse of power and thus reduces thepolitical costs of malfeasance.

An immediate corollary of the above proposition is that bridging social capital andcivic culture also improve private sector welfare WS = W0 − wC1(D)− (1−w)C0(D),which monotonically decreases in D and hence increases in a.

The dependence of private sector welfare on the bonding social capital measurew is not as straightforward due to the presences of the direct and indirect effectsdescribed earlier in this section. These effects work in opposite directions, and as aresult such dependence could be “non-linear”. The relative strength of the direct andindirect effects depend inter alia on the level of government accountability a, whichintegrates bridging social capital and civic culture. When such features of society areabsent, the government is completely unaccountable (a = 0), sets its expropriationat the highest possible level D = D, and the indirect effect thus disappears. In themeantime the direct effect of private protection from rampant government abuse couldbe substantial, and therefore the overall returns to bonding social capital in the absence(or near absence) of bridging social capital and civic culture should be expected to be

9

(perhaps, mildly) positive – in this case bonding social capital serves as an imperfectsubstitute for the bridging one. At the opposite extreme of full accountability (a = 1)which corresponds to very high stocks of bridging social capital and civic culture, thegovernment refrains from expropriation (D = 0), and therefore there is no need forprivate protection, and bonding social capital is idled. The returns to bonding socialcapital in this case should be zero.

Various specifications of the above model presented in Appendix A show that forintermediate levels of the accountability the indirect effect could be stronger thanthe direct one. In this case the substitution between various kinds of social capitaldisappears and the returns to bonding social capital become negative not only for thequality of governance, but for the private sector welfare, too – bonding social capitalis still helpful “ex post”, for a given level of abuse by the government, but causes fargreater damage “ex ante” by increasing the scale of such abuse.3

The presented analysis summarizes in the following hypotheses.

1. bridging social capital and civic culture have positive impact on governmentperformance and social welfare.

2. bonding social capital adversely affects government performance; its impact onsocial welfare could be positive at very low levels of bridging social capital andcivic culture, becomes (increasingly) negative as bridging social capital and civicculture grow bigger, and goes back to zero for very high levels of bridging socialcapital ensuring full government accountability.

We now turn to empirically testing these hypotheses and measuring economic payoffto various kinds of social capital and cohesion in Russian cities.

4 Data

Our main source of data was an all-Russia survey conducted in September 2007 by thePublic Opinion Foundation (Fond Obshchestvennoe Mnenie) as part of the ongoingGeoRating polling program which covers a broad range of economic, social, politicaland cultural issues. The survey sample comprised 34,038 adult respondents from 1924cities, towns and villages located in 68 Russian regions; in each covered region thesample was representative and included at least 500 respondents.

The survey questionnaire comprised three clusters of questions: (i) on respondents’views, norms and values – answers to such questions are commonly used in social capitalmeasurement; (ii) on respondents’ satisfaction with economic and social conditions in

3Such working of bonding social capital is somewhat similar to economic consequences of corruptionwhich helps individuals and businesses to navigate through excessive administrative barriers, butmotivates the bureaucracy to raise such barriers in the first instance (Rose-Ackerman, 1999).

10

their places of residence, and on their assessment of accountability and performance oflocal governments; and (iii) on individual characteristics of respondents. The first andsecond groups of questions were used to calculate resp. independent (explanatory) anddependent variables, whereas the third group supplied control variable; the latter alsoincluded size and administrative status (national capital, regional capital etc.) of thecity.

The first group of questions resembles (and at times replicates) those used in theWorld Values Survey and similar international polls (Table 1); these questions revealrespondents’ perception of the cohesion, self-help and propensity for collective actionin the surrounding society. Other questions from the same group characterize respon-dents’ own norms, views and practices, such as trust, help to others, willingness tojoin collective action, and the sense of responsibility for the situation in respondents’families, local communities, and cities (towns, villages).4

Respondents’ satisfaction with their lives was used to proxy economic outcomes; noother reliable data that would serve this purpose were immediately available at the citylevel. Government effectiveness and accountability assessed by respondents (answersto the question “Do you think authorities understand and take into account interestsof people like you?”) plays a dual role in the study – on the one hand accountablegovernance is of independent value of its own and thus an important outcome of socialcapital (Putnam, 1993; Knack, 2002); on the other hand government performanceis a plausible link between social capital and economic outcomes through a verticaltransmission channel.

Individual characteristics of respondents included age, gender and ethnic origin(the latter were found insignificant in our regression analyses), education, income andself-assessed material welfare. Control variables also included size and administrativestatus of the city (settlement) – predictably, those were strongly correlated with incomeand welfare of residents (Table 2).

An important decision in choosing our empirical identification strategy was to selectan appropriate territorial entity to establish links between social capital and economicoutcomes. Social capital by definition is a community resource5, and communities areoften proxied, for a lack of better practical options, by some territorial boundaries.In studies of economic payoff to social capital for other countries the territorial unitsconsidered as social capital reservoirs were usually regions (US states (Knack, 2002),German Lander (Blume, Sack, 2005), provinces etc. elsewhere in the world (Tabellini,

4We did not use data on philanthropy and associational membership as possible sources of socialcapital indexes – philanthropy in rudimentary in contemporary Russia, whereas reported associationmembership is often fictitious or purely nominal. It is noteworthy that in a number of studies seekingto measure economic payoff to social capital associational membership did not have a significantimpact on economic performance and government efficiency (see e.g. Knack, Keefer, 1997).

5Perhaps at the cost of slight abuse of terminology, one could still talk about individual socialcapital that characterizes trust and trustworthiness of a person, her internalization of pro-social values,as well as participation in various social networks (Glaeser, Laibson, Sacerdote, 2002; Halpern, 2005).

11

2008)). In the present study we opted instead for the city (town, village) level ofanalysis. This choice was due to profound intra-regional variations of social valuesand norms, as revealed by our data (see also Petrov et al., 2010) which override theweaker sense of regional cultural identity. With such variations, potentially valuableinformation would be lost if regional averages were used. Besides, GeoRating datadid not include performance assessment for regional administrations. The downside ofstudying the economic impact of social capital at the city level is a dearth of social andeconomic statistical data that would complement (and verify) respondents’ subjectiveassessment of social and economic conditions in their cities – urban statistical data inRussia are much more scarce than those collected for regions.

Within cities and towns cultural attitudes are more homogeneous, but exhibit sig-nificant inter-city variations across the sample – standard deviations could be as highas 45% of the sample average. This means that there are, simply put, noticeably moreand less civic cities, towns and localities in Russia. Among large cities (with popula-tion 100,000 and more) such variations are somewhat less pronounced, but still quiteperceptible (Table 3). Furthermore, local governments’ performance and residents’satisfaction with conditions in their cities fluctuate within broad margins, too. Theobserved variations present a “natural experiment” that bodes well for measuring theimpact of social capital for social and economic outcomes at the city level.

Some of the respondents’ values and attitudes are significantly correlated with eachother (Table 4) – these correlations could be evidences of more general latent featuresthat underlie reported norms and behavior. Making such features explicit is importantfrom substantive and instrumental points of view. Substantively, this could revealparticular types and patterns of social capital relevant for socio-economic outcomes;instrumentally, it would prevent multicollinearity in regression analysis.

Factor analysis of our data indeed produces three dominant factors (Table 5). Thefirst factor aggregates with high positive weights features of broad social cohesion, ac-cord, mutual help and propensity for collective action. Trust also enters into this factor,although with somewhat smaller coefficient. Overall, the first factor characterizes thecapacity for collective action within broad societal coalitions (‘Putnam groups’), andcan therefore be interpreted as a measure of bridging social capital.

The second factor integrates with highest loads the indexes of restricted and ex-clusive social connectedness and limited embeddedness of trust and pro-social norms(trust only in those who have much in common with a respondent, and preference todealing with such people). Indexes measuring broad social cohesion and propensity forcollective action enter the second factor with significant negative coefficients, reflectingcautious and possibly adverse attitude to ‘aliens’. These are the reasons to interpretthe second factor as an index of bonding social capital, which by definition facilitatesthe formation of exclusive ‘Olson groups’ providing club goods for their members,rather than working for common good at the society at large. Sensing threat to theirwell-being or shortage of essential resources and services, ‘Olson groups’ are mobilized

12

to alleviate such threats or provide necessary resources for their members internally,within the groups’ confines.

Finally, the third factor is positively linked with the sense of responsibility for whatis happening in the community and in the city. Such perception reflects awareness ofcitizen’s rights and duties, and can be interpreted as an index of civic culture.

The proposed interpretation of these three factors is somewhat imprecise and sub-ject to caveats (common in the social capital literature), but by and large it agreeswith the prevailing understanding and perception of the above concepts. The obtainedaggregation is robust: alternative factor analysis techniques produce similar results.6

The three types of social capital are significantly correlated with individual char-acteristics of respondents (Table 7); in particular bridging social capital is positivelycorrelated with education, income, and material well-being. Positive contribution ofeducation in accumulation of social capital is a well-established fact (see e.g. Gaeseret al, 2004), which has also been recently confirmed empirically for Russia (Natkhov,2010). bonding social capital, on the contrary, is more prevalent among less educatedand less economically successful groups. It is noteworthy that bridging social capitalis positively, and bonding – negatively associated with respondents’ age; this could bedue to the damage caused to the social capital of older cohorts by the two decades oftumultuous economic transition (Aghion et al., 2010). Finally, civic culture decreaseswith income (perhaps this reflects greater satisfaction with the status quo and higherprivate costs of civic activism to wealthier individuals), as well as with the size andstatus of the city – in large megapolises there could be stronger sense of alienation frompublic affairs and feeling of impossibility to influence public decision-making.

The capital city of Moscow (where the survey sample is representative) is a case inpoint. The stock of bridging social capital in Moscow is close to the national average,whereas bonding social capital is above, and civic culture – well below their averagelevels. The average level of bridging social capital is sustained by education of Moscowresidents, lack of civic culture is due to higher incomes, whereas higher stocks of bondingsocial capital can be explained by abnormal concentration of wealth and significantsocial and economic inequality which adversely affects trust and breeds rivalries andrent-seeking.

In what follows social capital indexes are normalized so that their minimal valuesare zero, and standard deviations equal unity.

6Our procedure is somewhat similar to Bjørnskov’s (2006), where factor analysis also producedthree orthogonal aggregate indexes of social capital reflecting resp. associational activities, socialnorms and trust. However in Bjørnskov (op. cit.) the set of original characteristics aggregated byfactor analysis was pre-selected “around” the above triplet, and hence the obtained aggregation wasless “endogenous” than in our case where no structure was assumed a priori and the interpretation ofthe obtained aggregates is based solely on the loads of primary characteristics.

13

5 Social capital and development: an empirical anal-

ysis

In most of the studies reviewed in Section 2 the payoff to social capital is measured interms of quality of governance or various socio-economic outcomes. In our regressionswe follow both of these traditions. In the first case the dependent variable (hereafterOutcome) is produced by averages of respondents’ assessments of socio-economic con-ditions in their cities and other types of settlements. This variable is regressed on thethree indexes of social capital – open, bonding, and civic culture (resp. BridgingSC,BondingSC, and CivicCulture), which are also averaged across the same localities. Inthe second case the dependent variable is the average of respondents’ assessment of theperformance of their local governments (Performance); however such variable can alsoserve as an explanatory one, to assess the contribution of governance to local devel-opment and investigate the role of governance as a transmission mechanism betweensocial capital and economic outcomes. Control variables are cities’ size, status, regionaldummies, and the averages of various individual characteristics of respondents.

The first regression model estimates the contribution of social capital to local de-velopment and welfare:

Outcomei = const+ β1BridgingSCi + β2BondingSCi + β3CivicCulture+

+ γiControlki +RegionDummyi + εi (4)

Here i is a settlement index. OLS estimation of this model (Table 8 , column 1)provides strong support to the hypothesis that bridging social capital makes positivecontribution to development: the corresponding coefficient is significant at the 1%level and quite substantial: a one standard deviation change in the bridging socialcapital index is associated with improvement of social and economic conditions in thecity by quarter of a standard deviation. The contribution of bonding social capitalis highly significant, too, but negative. Finally, the contribution of civic culture is,similarly to bridging social capital, positive (and significant at the 0.05 level), but oflesser magnitude. The obtained estimation is fully consistent with the predictions ofthe theory presented in Appendix A.

To check robustness, we include in the regression various controls; such modifica-tions leave estimated coefficients and their significances practically intact (columns 2-4of Table 8). In particular, size of the city and material well-being of residents have theexpected positive impact on the dependent variable, but the inclusion of these controlsdoes not affect the magnitude and significance of social capital contributions. Overallwe can conclude that more civic among Russian cities and towns enjoy ceteris paribusgreater prosperity and higher well-being.

14

In the second regression model the dependent variable is government performance:

Performancei = const+ β1BridgingSCi + β2BondingSCi + β3CivicCulture+

+ γkControlki +RegionDummyi + εi (5)

Estimations of the above model (Table 9) show that social capital’s impact on thequality of local governance is essentially the same as for social and economic outcomes –government effectiveness is positively and highly significantly associated with bridgingsocial capital and civic culture, and also highly significantly, but negatively – withbonding social capital. Here again the hypotheses generated by the theoretical modelfind full confirmation in the data. In the regression model (5) the association of thedependent variable with social capital is even stronger than in (4): the correspondingcoefficients have larger absolute values. These conclusions are also robust to variationsin the composition of control variables (columns 2 and 3 and the table). Moreover,they remain qualitatively unchanged if the full sample is reduced only to larger cities(100,000 residents and up; columns 4-7), and the impact of social capital on governmentperformance for such sub-sample becomes even stronger: one standard deviation inthe bridging social capital corresponds to one standard deviation of the quality ofgovernance index. Figure 2 illustrates this close association between social capital andthe quality of urban governance. These are evidences that more civic of the Russiancities and towns with broader social cohesion and stronger civic culture are by andlarge better governed.

Finally, in the third model social and economic outcomes – the dependent variable– are regressed on government performance; the set of dependent variables in suchmodel can also include indexes of social capital (Table 10, resp. column 1 and columns2-4):

Outcomei = const+ β0Performancei

+ β1BridgingSCi + β2BondingSCi + β3CivicCulture

+ γkControlki +RegionDummyi + εi (6)

The first column of Table 10 shows that the quality of local governance is highlysignificant for social and economic outcomes. In combination with OLS estimations ofthe equation (5) which demonstrate the significance of social capital for the quality ofgovernance, we can now conclude that the data point out to the working of a verticaltransmission channel between social capital and development, and local governmentsare the linchpin of such channel. This channel carries up to 50% of the contribution ofsocial capital to development: when social capital indexes are included in the regres-sion alongside the government performance index, the coefficients with such indexes(reflecting the horizontal channel) decrease almost by half in comparison with the re-gression model (4). In large cities the vertical channel becomes predominant and the

15

horizontal one nearly disappears – for such sub-sample social capital coefficients in themodel (6) become insignificant.

The above empirical models also shed light on the interplay between different typesof social capital in affecting local development and governance. In particular the theorypresented in Section 3 suggests that the adverse impact of the bonding social capitalgrows stronger as the stock of the bridging social capital increases in a low-to-mediumrange. We test this conjecture by dividing the sample in three parts with lower, interim,and higher stocks of the bridging social capital, and estimating the regression model(4) separately for each part. The results are reported in Table 11 which shows that thecoefficient of the bonding social capital is insignificant (and small) for the lower portion,and then becomes negative, significant at the 1% level and growing in magnitude as thebridging social capital increases from the middle to the top third of the sample. Theseestimations concur with the theoretical model (and its specifications in the AppendixA): the first portion of the sample corresponds to the low range of bridging social capitalwhere the returns to bonding social capital turns from initially positive to negative;such returns remain negative and increasing in magnitude thereafter (apparently thestock of open social capital in our sample does not reach the level when bonding socialcapital starts losing its significance.) We can therefore conclude that bonding socialcapital becomes increasingly a drag on local development when civic awareness andcapacity for collective action grow stronger.

6 Validation and causality

Validity of the above findings and conclusions could be questioned due to possibleomitted variable bias, measurement errors in data collection, and reverse causality.Control variables included in the regression models, and various robustness checks dealwith the omitted variable bias. Concerns about the quality of measurement could beraised inter alia due to the fact that almost all of our data come from a single surveyand are thus susceptible to sampling and polling errors. To address such concerns, wehave performed external validation by using similar data from other sources.

The proxy for social and economic outcomes – respondents’ satisfaction with situa-tions in their cities – was validated by data from other Geo-Rating surveys conductedbefore (2005) and after (2008, 2009) the 2007 poll. In those surveys respondents wereasked about satisfaction with social and economic conditions in their regions, ratherthan cities, and therefore such surveys’ data are only partially compatible with the2007 poll, but still allow for meaningful cross-checking (Table 12).

Replacing outcome measures by those similarly derived from other years’ surveysdo not qualitatively change the conclusions about the role of bridging and bondingsocial capital and civic culture in urban and municipal development.

Our search for external validation of the quality of governance and accountability

16

measures is still work in progress. Electoral statistics to which one would normally turnfor measures of political competition and other proxies for government accountabilityhas not been very useful so far, possibly due to massive irregularities in Russian localelections.

We have similarly performed validation of social capital indexes by using measures ofsocial accord and cohesion derived from a recent 2009 GeoRating survey; the obtainedresults were close to those reported in the previous section. We were unable to rely onblood donation and referenda participation data which are often used in social capitalmeasurement due to concerns about their accuracy and adequacy of such measures inRussia (e.g. much of blood donation in the country is motivated by material rewards).

Finally, we turn to the endogeneity problem in the association between social capitaland economic outcomes. One can argue that social capital is not only a factor, butalso a product, of development; one possible explanation of such reverse causality isthat development expands and improves education which is known to be a powerfuldriver of social capital accumulation. Similarly good governance could instill greatertrust in institutions and broader cooperation that would also be conducive for socialcapital buildup.

To be able to argue that social capital affects development, we need valid instru-ments for the social capital indexes. Features that were used as instruments for socialcapital elsewhere in the literature (see Section 2) in our case either failed the validitytest, or no satisfactory data and/or measures for such potential instruments were found.We had more luck with using the size of the middle class as a potential instrument.

Middle class is known to be conducive for the cultivation of civic values and cohe-sion (see e.g. Moore, 1966; Hooghe, Stolle, 2003; Easterly, Ritzen, Woolcock, 2006),and as such could indeed serve as a potentially valid instrument for social capital.Among multiple sources of information on middle class which reflect various measuresand interpretations of this broad concept, we have selected, based on availability andreliability of data, a survey conducted in 1980 by the Institute of Sociological Stud-ies of the Soviet Academy of Science (Levyikin et. al., 1980) that was comparable inits scale, scope and methodology to the 2007 Geo-Rating survey. The survey did notspecify cities, but available information on regions and city types enabled us to collecta sub-sample including 52 cities and towns. While such sample falls short of whatis ideally required for instrumental variable analysis, it still produces a satisfactoryinstrument for the 2007 index of cohesion and accord. The middle class proxy thatwas used to obtain the instrument was respondents’ description of their social status(sluzhashchie – professionals, white collars, etc., as opposed to workers and peasants).Two-stage least squares estimation shows that our proxy for the middle class in 1980is indeed a valid instrument for social capital in today’s Russia (see also Figure 3).The availability of such instrument lends some support to the causality that runs fromsocial capital to development.

17

7 Conclusion

We have shown that social capital does have substantial economic payoff in Russia,despite serious reasons to expect otherwise. It means that Russia, being a ‘normalcountry’, can rely on its social capital as a development resource.

This conclusion however is subject to an important qualification: while some kindsof Russian social capital advance development, others obstruct it. Rose (1998, p. 18)pointed out to a path dependency in Russian social capital and cohesion which ‘en-couraged people to create informal networks as protection against the state’. Suchbonding forms of social capital are considered ‘anti-modern’, as opposed to modernones, which ensure accountable governance and uphold economic, legal and politicalinstitutions (Polishchuk 2010). Our analysis demonstrates that in today’s Russia mod-ern and anti-modern types of social capital co-exist in proportions that vary from onecity and region to the other and likely evolve over time. It means that the agenda ofRussian modernization, apart from its technological and institutional aspects, has animportant social dimension, and that the evolution of the social capital mix could havefar-reaching implications for the nation’s economic and political development.

A sanguine development view holds that economic growth and accumulation ofhuman capital foster civic culture and pro-social values (Glaeser, Ponzetto, Shleifer,2007), which in their turn improve institutions and governance in the economy andsociety (Glaeser et al., 2004). On the other hand bonding social capital could disruptthis dynamic virtuous circle by perpetuating ineffective and unaccountable governanceand debasing modern institutions. Corruption, lawlessness and government preda-tion erode trust in institutions and among individuals, and suppress investments inbridging social capital and cultural transmission of pro-social norms and civic virtues(Tabellini, 2008), while entrenching anti-modern social practices of adjustment to badinstitutions.7

The established link between municipal government performance and social capitalmix could also involve a reverse causality, when the proportion between bridging andbonding social capital reflects efficiency, fairness and transparency (or lack thereof)in social services delivery (Kumlin, Rothstein, 2005). Corruption and injustice ingovernment undermine social trust (and hence bridging social capital) and at the sametime mobilize bonding social capital to seek “parochial” grassroots protection withinsmaller groups from government failures and abuse.

The outcome of such “race” between different kinds of social capital is uncertain,and multiple equilibria are possible. Further research, theoretical and empirical, isrequired to get a better insight into the processes of accumulation and amortization ofdifferent kinds of social capital. Such insight would be invaluable in designing policies

7“If you expect to live in a corrupt society, you would rather learn to pay and demand bribes”(Aghion et al., 2010,, p. 1027)

18

that would tip the race between modern and anti-modern social capital and cohe-sion towards a path where civil society, economic development, and good governancesupport and reinforce each other.

19

References

[1] Acemoglu, D, and J. Robinson. (2008) Persistence of Power, Elites, and In-stitutions. American Economic Review, 98, 267-293.

[2] Aghion, P., Y. Algan, P. Cahuc, and A. Shleifer. (2010) Regulation and Dis-trust. Quarterly Journal of Economics, 125, 1015-1049

[3] Almond, G., and S. Verba (1963). The Civic Culture. Political Attitudes andDemocracy in Five Nations. Princeton: Princeton Univ. Press.

[4] Arrow, K. (1972). Gifts and exchanges, Philosophy and Public Affairs, 1, 343-362. Bjornskov, Ch. (2006) The multiple facets of social capital. EuropeanJournal of Political Economy, 22, 22-40

[5] Blume L., and D. Sack. (2008). Patterns of Social Capital in West GermanRegions. European Urban and Regional Studies, 15, 229-248.

[6] Casey T. (2004) Social capital and regional economies in Britain. PoliticalStudies, 52, 96-117.

[7] Dasgupta, P. (2003) Social capital and economic performance: Analytics. In:E. Ostrom and T. Ahn, Editors, Foundations of Social Capital, Edward Elgar,Cheltenham, 309-339.

[8] Durlauf S. (2002) On The Empirics Of Social Capital Economic Journal, 112,459-479.

[9] Durlauf, S., and M. Fafchamp. (2005) Social Capital. In: Ph. Aghion, S.Durlauf (ed.), Handbook of Economic Growth, Vol.1, 1639-1699, Amsterdam:Elsevier.

[10] Dzialek, J. (2009) Social Capital and Economic Growth in Polish Regions.MPRA Working paper 18287.

[11] Easterly, W., J. Ritzen, and M. Woolcock (2006). Social Cohesion, Institu-tions, and Growth. Economics and Politics, 18, 103-120.

[12] Eberstadt, N. (2010). Russia’s Peacetime Demographic Crisis: Dimensions,Causes, Implications. NBR Project Report.

[13] Fukuyama, F. (1997). Social capital. Tanner Lecture on Human Value

[14] Glaeser, E., D. Laibson, J. Scheinkman, and Ch. Soutter. (2000) MeasuringTrust. Quarterly Journal of Economics, 115, 811-846.

20

[15] Glaeser, E., D. Laibson and B. Sacerdote. (2002) An Economic Approach toSocial Capital. Economic Journal, 112, 437-458.

[16] Glaeser, E., R. La Porta, F. De Silanes, and A. Shleifer. (2004). Do Institu-tions Cause Growth? Journal of Economic Growth, 9, 271-303.

[17] Glaeser, E., Ponzetto, G., and Shleifer, A. (2007) Why Does Democracy NeedEducation? Journal of Economic Growth, 12, 77-99.

[18] Grossman, G., and E. Helpman (2001). Special Interests Politics, CambridgeMA and London UK: The MIT Press.

[19] Guiso, L., P. Sapienza, and L. Zingales. (2010) Civic Capital as the MissingLink. NBER Working Paper No. w15845.

[20] Halpern, D. (2005). Social Capital. Cambridge: Polity.

[21] Hooghe, M., and D. Stolle (eds.) (2003) Generating Social Capital. CivilSociety and Institutions in Comparative Perspective. New York: PalgraveMacmillan.

[22] Knack, S. Social Capital and the Quality of Government: Evidence From theU.S. States. (2002) American Journal of Political Science 46, 772-785.

[23] Knack, S., and Ph. Keefer. (1997) Does Social Capital Have an Economic Pay-off? A Cross-Country Investigation. Quarterly Journal of Economics, 1251-1288.

[24] Kennedy, B., and I. Kawachi. (1998) The Role of Social Capital in the RussianMortality Crisis. World Development, 26), 2029-2043.

[25] Kumlin, S., and B. Rothstein. (2005) Making and Breaking Social Capital.The Impact of Welfare-State Institutions. Comparative Political Studies, 38,339-365.

[26] Kuran, T. (1991) The East European Revolution of 1989: Is it surprising thatwe were surprised? American Economic Review, 81, 121-125.

[27] La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R. Vishny. (1997) Trustin Large Organizations. American Economic Review, 87, 333-338.

[28] Levykin I., T Dridze, E Orlov, and Y. Reyzema. (1981) State and Main Trendsof the Soviet Way of Life. (In Russian) Materials of a Public Opinion Poll.

[29] Marsh, Ch. (2000) Social Capital and Democracy in Russia. Communist andPost-Communist Studies, 33, 183-199.

21

[30] Moore, B. (1966) Social Origins of Dictatorship and Democracy, Boston; Bea-con Press.

[31] Natkhov, T. (2010). Education and Social Cohesion in Russia. An EmpiricalEvidence. Mimeo.

[32] Ostrom, E. (2000) Social capital: A fad or fundamental concept?. In: P. Das-gupta and I. Seragilden, Editors, Social Capital: A Multifaceted Perspective,World Bank, Washington, DC, 172-214.

[33] Petrov N., and A Titkov A. (2010) Irregular Triangle: State-Business-SocietyRelations in Russia?s Regions (in Russian) Moscow: Rosspen.

[34] Polishchuk, L. (2010) Misuse of Institutions. Lessons from Transition. UNU-WIDER Working Paper No 2010/75.

[35] Polterovich, V (2007). Elements of Reform Theory (in Russian). Moscow:Economics Publishers.

[36] Putnam, R. (1993) Making Democracy Work: Civic Tradition in ModernItaly. Princeton: Princeton Univ. Press.

[37] Rodrik, D, A. Subramanian, and F. Trebbi. (2004) Institutions Rule: ThePrimacy of Institutions Over Geography and Integration in Economic Devel-opment, Journal of Economic Growth, 9, 131-165.

[38] Roland, G. (2000) Transition and Economics. Politics, Markets, and Firms.MIT Press.

[39] Rose, R. (1998) Getting Things Done in an Anti-Modern Society: SocialCapital Networks in Russia. Social Capital Initiative Working Paper No. 6.World Bank.

[40] Rose-Ackerman, S. (1999) Corruption and Government: Causes, conse-quences,and reform. Cambridge: Cambridge University Press.

[41] Rupasingha A, S. Goetz, and D. Freshwater. (2002) Social and InstitutionalFactors as Determinants of Economic Growth: Evidence from the UnitedStates Counties Papers in Regional Science 81, 139-155.

[42] Shleifer, A., and D. Treisman. A Normal Country: Russia After Communism.(2005). Journal of Economic Perspectives, 19, 151-174.

[43] Tabellini, G. (2008) Institutions and Culture. Journal of the European Eco-nomic Association 6, 255-294.

22

[44] Weingast, B. (1997). The Political Foundations of Democracy and the Ruleof Law. American Political Science Review, 91, 245-263.

[45] Wolfe, A. (1989). Whose Keeper? Social Science and Moral Obligation. Berke-ley: University of California Press.

[46] Woolcock, M. Social Capital and Economic Development: Toward a Theoret-ical Synthesis and Policy Framework. Theory and Society, 27, 151-208.

[47] Woolcock, M., and D. Narayan. (2000). Social Capital: Implications for De-velopment Theory, Research, and Policy. World Bank Research Observer, 15,225-249.

[48] Zak, P. and S. Knack. (2001). Trust and Growth. Economic Journal, 111,295-321.

[49] Zhang Zh., and R. Ke . (2003) Trust in China: A Cross-Regional Analysis,William Davidson Institute Working Paper No. 586.

23

A Model Specifications

We present two specifications of the general model described in Section 3. In bothversions government resorts to distortionary tools of income extraction that imposeexcess burden on the private sector. Bonding social capital allows members of smallgroups to eliminate the excess burden by means of self-organization and reduce lossesfrom C0(D) down to C1(D) = D.

Specification 1: Extortionary taxation Suppose that the government extractsincome from the private sector through an extortionary tax with flat rate t ∈ [0, 1].Assuming agents’ quasilinear utilities x− v(l), where x is income and l – labor, laborsupply l = l(t) can be found from the equation v′(l) = 1−t (market wage is normalizedto unity), and government revenue is D = R(t) ≡ tl(t). Residual welfare of a taxpayerafter taxes is V (t) ≡ (1 − t)l(t) − v(l(t)), and when there is no grassroots protectionfrom taxation, the cost of government predation to agents is as follows: C(D) =V (0)− V (t) = R(t) + L(t), where L(t) is the deadweight loss of a distortionary tax.

Bonding social capital enables agents within small groups to accumulate the re-quired tax payment per member trough direct contributions without sustaining thedeadweight losses8; the saved deadweight losses comprise economic returns (which ac-crue to group members) to bonding social capital. In this case C1(D) = D, and thetax rate t = t(a, w) selected by the government from problem (3) satisfies the followingfirst-order condition:

1− aa

R′(t) = (1− w)L′(t).

The social welfare as a function of a and w is as follows:

WS(a, w) = W0 −R(t(a, w))− (1− w)L(t(a, w)).

This function is increasing in a, and one can easily check that under the “neoclassical”assumptions about v(·), decreases in w for large enough a. This is illustrated by theprofiles of WS(a, w) and D(a, w) for v(l) = l2 (Figure 1a below).

Specification 2: Diversion of public funds Assume agents’ preferences of theform x + f(G), where x is private consumption, and G – local public good, witha “neoclassical” function f . Local public goods are supposed to be provided by thegovernment for N identical communities of equal size 1/N (consumers are still assumedto form a unit continuum). Optimal provision G = G∗ of the local public good for eachcommunity can be found from the equation

f ′(G∗) = N

8Such outcome obtains e.g. as political equilibrium when agents’ groups are lobbies making con-tributions to government in order to prevent taxation of group members (Grossman, Helpman, 2001)

24

Suppose that the government collects the required revenues NG∗, but can divertportion D of this amount for its own enrichment, leaving the public goods undersup-plied. In this case (assuming equal (under)funding of each of the local public goods)the cost to the agents of such diversion is C0(D) = f(G∗)− f(G∗ −D/N)

Bonding social capital could help agents within a given community to resolve thecollective action problem and make up for the shortfall of funding of the local publicgood by jointly supplying the missing amount D/N ; in such case each member ofthe community will have to make a private contribution D, and the private cost ofgovernment malfeasance is reduced from C0(D) to C1(D) = D. The stock of bondingsocial capital is measured by the share w of the communities where such local effortoccurs; in this case problem (3) takes the following form:

maxD{D − a[wD + (1− w)(f(G∗)− f(G∗ −D/N))]}

and the optimal diversion of funds D = D(a, w) satisfies the equation

f ′(G∗ −D) = N1− awa(1− w)

Here too D(a, w) monotonically decreases in a and increases in w, and the socialwelfare is as follows:

WS(a, w) = W0 − wD(a, w) + (1− w)(f(G∗)− f(G∗ −D(a, w)/N)

As it was the case with the previous specification, with the “neoclassical” assump-tions this function can also be shown to decrease in w at least when a is sufficientlylarge. We illustrate this by the profile of WS(a, w) and D(a, w) for f(G) =

√G (Figure

1b below).

25

Fig

ure

1:P

rofile

sofD

(a,w

)an

dW

S(a,w

)fo

rco

nsi

der

edsp

ecifi

cati

ons:

(a)

forv(l

)=

l2(s

pec

ifica

tion

1)

(b)

forf

(G)

=√G

(sp

ecifi

cati

on

2)

26

BE

mpir

ical

Analy

sis

Resu

lts

Tab

le1:

Soci

alnor

ms,

atti

tudes

,an

dou

tcom

es

Var

iab

leQ

ues

tion

Maxa

unit

How

ofte

np

eop

lear

oun

dyo

uar

ep

rep

are

dfo

rco

llec

tive

act

ion

tojo

intl

yso

lve

thei

rp

rob

lem

s?4

unit

self

How

ofte

np

eop

lear

oun

dyou

are

pre

pare

dfo

rco

llec

tive

act

ion

tojo

intl

yso

lve

soci

al

pro

ble

ms,

even

ifth

ela

tter

hav

en

oim

med

iate

bea

rin

gsfo

rth

em?

4

agrall

Inyo

ur

opin

ion

,w

hat

ism

ore

com

mon

inou

rco

untr

yto

day

soci

al

acc

ord

an

dco

hes

ion

,or

dis

cord

an

dal

ien

atio

n?

3

agrclose

Inyo

ur

opin

ion

,w

hat

ism

ore

com

mon

am

on

gp

eople

aro

un

dyo

uso

cial

acc

ord

an

dco

hes

ion

,or

dis

cord

an

dal

ien

atio

n?

3

trust

Do

you

thin

kth

atp

eop

leca

nb

etr

ust

ed,

or

you

can

not

be

more

care

ful

ind

eali

ng

wit

hp

eop

le?

1com

val

Do

you

mee

tp

eop

leth

ath

ave

mu

chin

com

mon

wit

hyo

u?

3trust

com

Do

you

tru

stth

ose

wh

oh

ave

mu

chin

com

mon

wit

hyou

more

,le

ss,

or

the

sam

eas

all

oth

ers?

3help

How

ofte

np

eop

lear

oun

dyo

uar

ere

ad

yto

hel

pea

choth

er?

3ind

unit

Som

ep

eop

lear

ere

ady

tojo

inoth

ers

for

join

tact

ion

on

lyif

they

hav

eth

esa

me

inte

rest

san

dsh

are

the

sam

eid

eas.

Oth

ers

are

read

yfo

rjo

int

act

ion

even

ifp

art

ner

sin

tere

sts

an

did

eas

are

diff

eren

t.T

ow

hic

hof

thes

etw

ogr

oup

syo

uar

ecl

oser

?

3

ind

help

Hav

eyou

over

the

last

yea

roff

ered

ass

ista

nce

an

dsu

pp

ort

toth

ose

wh

oare

not

you

rim

med

iate

fam

ily

mem

ber

s?3

resp

fam

How

stro

ngl

yyo

ufe

elre

spon

sib

ilit

yfo

rth

esi

tuati

on

inyo

ur

fam

ily?

3re

spoutd

rsH

owst

ron

gly

you

feel

resp

onsi

bil

ity

for

the

situ

ati

on

inyo

ur

apart

men

tb

uil

din

gor

loca

lre

sid

enti

alare

a?

3re

spcity

How

stro

ngl

yyo

ufe

elre

spon

sib

ilit

yfo

rth

esi

tuati

on

inyo

ur

city

(tow

n,

vil

lage)

?3

socpow

Do

you

thin

kau

thor

itie

su

nd

erst

an

dan

dta

kein

toacc

ou

nt

inte

rest

sof

peo

ple

like

you

?3

socoutc

Ove

rall

,ar

eyo

usa

tisfi

edor

dis

sati

sfied

by

the

situ

ati

on

inyo

ur

city

(tow

n,

vil

lage)

?3

aA

llan

swer

sar

en

orm

aliz

edso

that

thei

rm

inim

al

valu

esare

zero

;gre

ate

rva

lue

corr

esp

on

ds

tost

ron

ger

agre

emen

t,h

igh

erfr

equ

ency

etc.

27

Table 2: Distribution of Individual characteristics of respondents

Size and status of settle-

ment

Sample Averageage

Averageyears ofeduca-tion

Averagewelfare

Averageincome

Moscow 1 45 11.68 3.21 9.91

St. Petersburg 1 44.54 11.2 3.22 10.55

Regional capital with more

than 1,000,000 residents

11 43.8 10.72 3 5.86

Regional capital with less than

1,000,000 residents

56 44.2 10.82 2.93 5.54

Towns, small urban settlements 909 44.46 10.24 2.74 4.75

Villages 844 46.9 9.44 2.49 3.59

Total 1822 45.58 9.89 2.63 4.25a Respondents were asked to estimate their material welfare in a one (“not enough money even for

food”) to six (“experience no financial difficulties, could buy a house or apartment if need be”)scale. Income was reported in thousands of rubles.

Table 3: Variations of city averages across the sample and among larger cities

VariableFull Sample (Nobs = 1822) Large cities (Nobs = 149)

Mean Std.

Dev.

Min Max Mean Std.

Dev.

Min Max

unit 1.67 0.53 0.00 3.40 1.74 0.36 0.41 2.80

unit self 1.23 0.56 0.00 3.30 1.23 0.43 0.08 2.41

agr all 0.84 0.38 0.00 2.80 0.82 0.29 0.17 1.60

agr close 1.64 0.45 0.10 3.00 1.76 0.31 1.00 2.82

trust 0.19 0.16 0.00 1.00 0.18 0.10 0.00 0.50

com val 2.02 0.34 0.68 3.00 2.03 0.24 1.10 2.63

trust com 1.92 0.42 0.17 3.00 1.96 0.31 1.20 2.80

help 1.86 0.46 0.30 3.30 1.94 0.31 0.56 2.88

ind unit 1.84 0.51 0.00 3.00 1.82 0.39 0.44 2.76

ind help 1.76 0.50 0.00 3.00 1.78 0.37 0.70 2.61

resp fam 2.71 0.26 1.29 3.00 2.68 0.16 2.00 3.00

resp outdrs 1.78 0.59 0.00 3.00 1.36 0.37 0.12 2.61

resp city 1.08 0.52 0.00 3.00 0.91 0.35 0.08 2.35

soc pow 0.89 0.38 0.00 2.38 0.89 0.30 0.20 1.84

soc outc 1.11 0.47 0.00 2.63 1.31 0.38 0.20 2.25

28

Tab

le4:

Pai

rwis

eco

rrel

atio

ns

ofso

cial

nor

ms

and

atti

tudes

un

itu

nit

self

agr

all

agr

close

tru

stco

mva

ltr

ust

com

hel

pin

du

nit

ind

hel

pre

spfa

mre

spou

tdrs

un

it1

un

itse

lf0.

681

agr

all

0.40

0.4

11

agr

clos

e0.

280.1

60.

341

tru

st0.

190.1

80.

350.

181

com

val

0.19

0.0

9-0

.01

0.26

0.16

1tr

ust

com

0.04

-0.0

5-0

.13

0.24

0.06

0.46

1h

elp

0.61

0.4

00.

370.

360.

230.

230.

151

ind

un

it0.

290.2

20.

110.

270.

110.

350.

280.

221

ind

hel

p0.

140.0

9-0

.02

0.16

-0.0

10.

300.

240.

280.

311

resp

fam

0.03

-0.0

1-0

.07

0.07

-0.0

50.

110.

150.

050.

160.

22

1re

spou

tdrs

0.01

0.0

70.

010.

05-0

.03

0.04

0.04

0.00

0.16

0.11

0.34

1re

spci

ty0.

160.2

00.

160.

080.

080.

05-0

.02

0.10

0.20

0.0

90.2

30.

57

29

Table 5: Factor analysis results. Proportion of factors invariation

Factor Eigenvalue Proportion Cumulative

Factor1 2.68 0.61 0.61Factor2 1.29 0.29 0.90Factor3 0.99 0.23 1.13Factor4 0.37 0.09 1.21Factor5 0.14 0.03 1.24Factor6 0.01 0.00 1.25Factor7 -0.01 0.00 1.25Factor8 -0.06 -0.01 1.23Factor9 -0.12 -0.03 1.2Factor10 -0.18 -0.04 1.16Factor11 -0.23 -0.05 1.11Factor12 -0.24 -0.05 1.06Factor13 -0.24 -0.06 1.00

Table 6: Factor analysis results. Factor loadings for the whole sample and for thesub-sample of large cities

VariableFull Sample Large cities

Fact1 Fact2 Fact3 Uniq. Fact1 Fact2 Fact3 Uniq.unit 0.74 -0.30 0.02 0.36 0.67 -0.39 -0.31 0.30unit self 0.61 -0.32 0.17 0.50 0.44 -0.46 -0.24 0.54agr all 0.49 -0.39 0.13 0.60 0.51 -0.43 -0.05 0.56agr close 0.48 0.05 -0.16 0.74 0.57 0.30 0.02 0.58trust 0.32 -0.15 -0.05 0.87 0.43 -0.16 -0.13 0.77trust com 0.25 0.43 -0.37 0.62 0.41 0.50 -0.09 0.58com val 0.40 0.35 -0.33 0.61 0.47 0.45 -0.17 0.55help 0.67 -0.14 -0.13 0.51 0.61 -0.05 -0.03 0.63ind unit 0.48 0.29 -0.08 0.68 0.59 0.23 -0.01 0.59ind help 0.34 0.34 -0.15 0.75 0.39 0.43 0.08 0.66resp fam 0.16 0.40 0.18 0.78 0.13 0.26 0.39 0.77resp outdrs 0.20 0.42 0.54 0.50 0.28 -0.14 0.74 0.35resp city 0.32 0.25 0.55 0.53 0.34 -0.27 0.68 0.35

30

Table 7: Correlations of social capital and individual characteristics

Factor 1 Factor 2 Factor 3 Age Education Well-beingFactor 1 1Factor 2 -0.03 1Factor 3 0.01 0.01 1Age -0.16 0.14 0.05 1Education 0.17 -0.06 -0.09 -0.33 1Well-being 0.18 -0.08 -0.11 -0.26 0.37 1Income 0.12 -0.08 -0.21 -0.20 0.36 0.47

Table 8: Regression of social and economic outcomes on social capital

VARIABLES (1) (2) (3) (4)Bridging SC 0.122*** 0.114*** 0.122*** 0.114***

(0.010) (0.012) (0.010) (0.014)Bonding SC -0.091*** -0.088*** -0.091*** -0.088***

(0.003) (0.002) (0.003) (0.015)Civic culture 0.019** 0.025** 0.021*** 0.025*

(0.006) (0.009) (0.005) (0.014)Population 0.0013*** 0.0010**

(0.0002) (0.0003)Age -0.003 -0.003

(0.005) (0.003)Education 0.002 0.003

(0.009) (0.012)Wellbeing 0.115*** 0.116***

(0.016) (0.029)City size dummy NO NO YES YESRegional effects YES YES YES YESObservations 1822 1822 1822 1822R-squared 0.267 0.282 0.266 0.280a Robust standard errors clustered at settlement type are in parenthesis.

***:1%, **: 5%, *: 10%. Age, education and wellbeing are an average age,years of education and wellbeing of respondents in the locality accordingly.Population proxy is a log of number of respondents.

31

Tab

le9:

Reg

ress

ion

ofgo

vern

men

tp

erfo

rman

ceon

soci

alca

pit

al

Tota

lsa

mple

Larg

eC

itie

sV

AR

IAB

LE

S(1

)(2

)(3

)(4

)(5

)(6

)(7

)B

ridgi

ng

SC

0.12

8***

0.12

4***

0.12

3***

0.20

5***

0.20

4***

0.18

3***

0.16

5***

(0.0

02)

(0.0

04)

(0.0

04)

(0.0

16)

(0.0

12)

(0.0

17)

(0.0

28)

Bon

din

gSC

-0.0

98**

*-0

.095

***

-0.0

95**

*-0

.136

***

-0.1

36**

*-0

.118

***

-0.1

31**

*(0

.004

)(0

.005

)(0

.005

)(0

.027

)(0

.027

)(0

.022

)(0

.022

)C

ivic

cult

ure

0.05

7**

0.06

0**

0.05

9**

0.10

1***

0.10

1***

0.11

6***

0.12

2***

(0.0

18)

(0.0

20)

(0.0

20)

(0.0

08)

(0.0

06)

(0.0

09)

(0.0

03)

Pop

ula

tion

-0.0

001

-0.0

02-0

.015

-0.0

22*

(0.0

00)

(0.0

12)

(0.0

11)

(0.0

09)

Age

-0.0

03*

-0.0

03*

0-0

.006

(0.0

01)

(0.0

01)

(0.0

06)

(0.0

09)

Educa

tion

-0.0

07**

*-0

.004

***

0.00

20.

025*

*(0

.001

)(0

.001

)(0

.026

)(0

.006

)W

ellb

eing

0.06

4***

0.06

5***

0.16

7***

0.16

2***

0.00

00.

000

(0.0

11)

(0.0

08)

Cit

ysi

zedum

my

NO

NO

YE

SN

ON

ON

ON

OR

egio

nal

effec

tsY

ES

YE

SY

ES

NO

NO

NO

NO

Obse

rvat

ions

1822

1822

1822

8686

8665

R-s

quar

ed0.

289

0.29

60.

297

0.52

10.

521

0.56

10.

505

aR

obu

stst

and

ard

erro

rscl

ust

ered

atse

ttle

men

tty

pe

are

inp

are

nth

esis

.***:1

%,

**:

5%

,*:

10%

.A

ge,

edu

cati

on

and

wel

lbei

ng

are

anav

erag

eag

e,yea

rsof

edu

cati

on

an

dw

ellb

ein

gof

resp

on

den

tsin

the

loca

lity

acc

ord

ingly

.P

opu

lati

onp

roxy

isa

log

ofnu

mb

erof

resp

on

den

ts.

32

Tab

le10

:Im

pac

tof

per

form

ance

and

soci

alca

pit

alfo

rso

cial

and

econ

omic

outc

omes

Tota

lsa

mple

Larg

eC

itie

sV

AR

IAB

LE

S(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)P

erfo

rman

ce0.

450*

**0.

352*

**0.

338*

**0.

338*

**0.

813*

**0.

813*

**0.

855*

**0.

972*

**(0

.022

)(0

.014

)(0

.015

)(0

.016

)(0

.059

)(0

.058

)(0

.176

)(0

.075

)B

ridgi

ng

SC

0.07

8***

0.07

2***

0.07

2***

-0.0

38-0

.036

-0.0

27-0

.054

(0.0

08)

(0.0

09)

(0.0

09)

(0.0

34)

(0.0

45)

(0.0

41)

(0.0

52)

Bon

din

gSC

-0.0

60**

*-0

.056

***

-0.0

56**

*0.

058*

*0.

059*

0.04

8*0.

04(0

.006

)(0

.002

)(0

.003

)(0

.021

)(0

.026

)(0

.017

)(0

.024

)C

ivic

cult

ure

-0.0

070.

005

0.00

5-0

.096

-0.0

96-0

.116

-0.2

07**

*(0

.005

)(0

.004

)(0

.002

)(0

.082

)(0

.084

)(0

.089

)(0

.020

)P

opula

tion

0.00

11**

*0.

003

0.02

70.

024

(0.0

00)

(0.0

36)

(0.0

36)

(0.0

71)

Age

-0.0

02-0

.001

-0.0

41**

*-0

.03

(0.0

05)

(0.0

05)

(0.0

08)

(0.0

14)

Educa

tion

0.00

50.

004

-0.1

23*

-0.1

59**

(0.0

10)

(0.0

08)

(0.0

46)

(0.0

41)

Wel

lbei

ng

0.09

4***

0.09

3***

-0.0

79-0

.244

**(0

.014

)(0

.014

)(0

.192

)(0

.065

)C

ity

size

dum

my

NO

NO

NO

YE

SN

ON

ON

ON

OR

egio

nal

effec

tsN

OY

ES

YE

SY

ES

NO

NO

NO

NO

Obse

rvat

ions

1822

1822

1822

1822

8686

8665

R-s

quar

ed0.

137

0.31

90.

336

0.33

50.

197

0.19

70.

267

0.29

2a

Rob

ust

stan

dar

der

rors

clust

ered

atse

ttle

men

tty

pe

are

inp

are

nth

esis

.***:1

%,

**:

5%

,*:

10%

.A

ge,

edu

cati

on

an

dw

ellb

ein

gar

ean

aver

age

age,

year

sof

edu

cati

onan

dw

ellb

ein

gof

resp

on

den

tsin

the

loca

lity

acc

ord

ingly

.P

op

ula

tion

pro

xy

isa

log

of

nu

mb

erof

resp

ond

ents

for

tota

lsa

mp

lean

dlo

gof

nu

mb

erof

citi

zen

sfo

rla

rge

citi

es.

33

Tab

le11

:R

egre

ssio

nof

outc

omes

onb

ondin

gso

cial

capit

alby

quan

tile

sof

bri

dgi

ng

soci

alca

pit

aldis

trib

uti

on

the

firs

tth

ird

the

seco

nd

thir

dth

ela

stth

ird

the

firs

tth

ird

the

seco

nd

thir

dth

ela

stth

ird

VA

RIA

BL

ES

(1)

(2)

(3)

(4)

(5)

(6)

Bri

dgi

ng

SC

0.11

8***

0.00

993

0.09

87**

*0.

122*

**0.

0125

0.09

93**

*(0

.031

9)(0

.072

0)(0

.031

6)(0

.032

0)(0

.072

1)(0

.031

8)B

ondin

gSC

-0.0

207

-0.0

553*

**-0

.133

***

-0.0

221

-0.0

512*

**-0

.134

***

(0.0

193)

(0.0

181)

(0.0

170)

(0.0

194)

(0.0

181)

(0.0

171)

Educa

tion

0.02

810.

0062

60.

0116

0.01

810.

0066

60.

0160

(0.0

337)

(0.0

330)

(0.0

338)

(0.0

349)

(0.0

344)

(0.0

351)

Wel

lbei

ng

0.16

0***

0.12

9***

0.09

29**

0.15

6***

0.12

8***

0.09

37**

(0.0

383)

(0.0

357)

(0.0

377)

(0.0

385)

(0.0

358)

(0.0

380)

Age

-0.0

0006

0.00

134

-0.0

0010

80.

0011

2-0

.000

218

-0.0

0057

0(0

.004

44)

(0.0

0449

)(0

.004

42)

(0.0

0456

)(0

.004

53)

(0.0

0447

)P

opula

tion

0.00

070.

0009

**0.

0005

-0.0

004

0.00

080.

0008

(0.0

009)

(0.0

004)

(0.0

007)

(0.0

02)

(0.0

01)

(0.0

02)

Cit

ysi

zedum

my

NO

NO

NO

YE

SY

ES

YE

SO

bse

rvat

ions

610

614

608

610

614

608

R-s

quar

ed0.

077

0.05

90.

136

0.08

40.

074

0.14

0a

Rob

ust

stan

dar

der

rors

clu

ster

edat

sett

lem

ent

typ

eare

inp

are

nth

esis

.***:1

%,

**:

5%

,*:

10%

.A

ge,

edu

cati

onan

dw

ellb

eing

are

anav

erage

age,

years

of

edu

cati

on

an

dw

ellb

ein

gof

resp

ond

ents

inth

elo

cali

tyac

cord

ingl

y.P

opu

lati

onp

roxy

isa

log

of

nu

mb

erof

resp

on

den

tsfo

rto

tal

sam

ple

an

dlo

gof

nu

mb

erof

citi

zen

sfo

rla

rge

citi

es.

34

Table 12: Correlations between respondents’ assessments of socio-economic condi-tions in their cities in 2007 and similar regional assessments for other years

Total sample(1816)

Larger cities(85)

Regionalcapitals (66)

2005 regional survey 0,22 0,60 0,652008 regional survey 0,34 0,73 0,792009 regional survey 0,21 0,62 0,70

Figure 2: Social capital and quality of governance in large cities. Partial regression plotbetween the performance and bridging social capital measures (controlling for averageincome, education and city type)

35

Figure 3: Partial regression plot between the size of middle class in 1980 and socialcohesion in 2007 (controlling for average income, education in 1980 and 2007 and citytype)

36